ABSTRACT With the increasing adoption of computed tomography (CT) as a screening tool for lung cancer, methods for identifying the small number of patients with malignant nodules from among the large number of patients with benign CT-detected nodules is a growing and urgent clinical need. We have targeted the problem of developing biomarkers for detecting malignant solid or part-solid nodules that are 6 ? 25 mm in diameter that are identified by screening at risk individuals or found incidentally in screen-eligible individuals. The ability to sensitively detect lung cancer in this clinical setting could reduce many of the potentially harmful consequences that currently arise from uncertainties about which of these indeterminate lung nodules require the most aggressive workup. The core of our approach is the integration of molecular biomarkers measured in non-invasively collected nasal brushes and plasma specimens together with complementary imaging and clinical markers. On the basis of our preliminary data, we will use total RNA sequencing of both large and small RNA to deeply characterize the cancer-associated airway-wide field of injury in nasal epithelium; exosome-derived plasma miRNA to capture information about tumor-associated products found in the circulation; and qualitative and quantitative imaging characteristics to capture information about the biology of the nodule and the local environment that would otherwise only be available through direct sampling. Further, we will be profiling these features in several unique cohorts of smokers with indeterminate nodules detected either incidentally or by screening that represent the clinical population in which most lung cancers are diagnosed. Our use of biorepositories that have been collected from the clinical settings in which the biomarker would ultimately be applied, utilizing a prospective-specimen-collection, retrospective-blinded- evaluation (PRoBE) design minimizes potential bias and improves applicability to the intended use population. A key aspect of our biomarker development plan is a two-staged feature selection process that will allow us to efficiently use patient cohorts to detect robustly cancer-associated molecular and imaging features that will then be used to construct integrated cancer predictive models. The performance and clinical utility of the resulting models will undergo preliminary validation studies at the end of the proposed studies. This will allow us to make a GO / NO-GO decision about whether they should be subsequently tested in larger validation trials based on a rigorous evaluation of their validity and also whether they represent progress toward our goal of shrinking the intermediate risk category, thereby improving the diagnostic workup of the large number of patients for whom there is currently considerable clinical uncertainty.